Abstract As older adults make up a growing share of the population, the prevalence of dementia will likely increase, but the magnitude of the increase will depend on the course of age-specific rates. Most projections of Alzheimer's disease and dementia assume that age-specific prevalence rates do not change and that population aging alone will drive dementia prevalence trends. There is some recent evidence, however, that age-specific prevalence may be declining Europe and possibly in the United States. This study will use data from the Health and Retirement Study to estimate the trajectory of dementia prevalence and to examine risk factors. Since 1992, the HRS has surveyed biennially a panel of Americans over age 50, including about 8000 over 70. Data collected include assessments of cognition, health, socioeconomic status, family background, and now genetic information and biomarkers. A sub-study in the early 2000s administered an in-depth cognitive assessment covering dementia to 856 HRS respondents, and now a new sub-study to begin in 2016 will assess dementia status in a subsample of about 3000. This study has five specific aims. First, it will estimate the age-adjusted prevalence of dementia from 1998 to 2016 in the population over age 70 based on an improved model of dementia probability, including the use of genetic information; it will also estimate age-adjusted trends in dementia incidence (onset) and dementia mortality. Second, it will analyze subpopulations such as by gender or by race/ethnicity to document differences in dementia risk by groups. It will test whether any such differences can be explained by differences in exposure to risk factors such as diabetes or education and whether group-specific trends in risk factors affect group-specific trends in dementia. Third, where risk factors are found to be associated with dementia prevalence, the study will seek to determine whether these associations reflect causality. Fourth, using the models of the probability of dementia, the evidence about causality, and forecasts of risk factors, the study will predict dementia prevalence for the population and for important subpopulations to 2040. Fifth, the study will forecast the use and availability of informal care, e.g., care by family members, and how it will affect nursing home use and costs.